> Through my recent explorations with Squeak, I observe that a 
> substantial amount of valuable Contrib code is being defacto jettisoned 
> version-to-version.  (rough empirical estimate: anything not in 
> -current has about a 15% chance of working).  This is both discouraging 
> for the newcomer and hinders growth.  Because modules do not stay 
> viable, peripheral contributors are re-implementing the same wheel 
> repeatedly.

This seems to be a substantial, perhaps wild, exaggeration.  Almost all 
of the image tends to survive version to version, so far as I can tell.

Maybe I'm just lucky, but II have code that has been used and working 
with the current version since version 2.4.  No doubt that substantial 
changes are introduced from time to time, and some code does get 
undercut by new stuff.  (No doubt a substantial amount of code did not 
survive the alignment morphs without some changes, and reasonable people 
may determine that the change wasn't worth it -- I think most would not 
agree.)  Likewise, I have some old, but well-written, Python code that 
wouldn't be workable under the new version without a complete rewrite.

Modularity does not equate to backward compatibility.

BTW,, based on Larry Wall's Apocalypse articles, a whole bunch of Perl 
5.0 code won't work in Perl 6 either.  Perhaps they ought to fork off a 
Stable Perl?

Recent arguments made by Microsoft executives criticizing open source 
development criticize such systems point to forking software as its 
achilles heel.

That being said, thanks for the contribution.  Here's hoping for many 
more.